Creating a positive IO - page 17

 
Kocty:

You as well as the theorists always cite ideal cases, if the system will work in the limit.... Of course, when it starts making money, whether one person or many, the market will just swallow it up and crush it, and it will be ruined. Why would you want to take everything out of the market? In this case, the market won't give a damn if the system crashes a bit, and you will live happily ever after with it.
Well, I'm not the only one who will. And when there's a lot of quickies like me, then... NIVELIATING :-)
 

We describe what we see on the screen. The description may be anything, as long as our imagination is good. e.g. indicator readings, interest rates, turnover, macroeconomic indicators, etc., taste, colour, etc. We shove the whole pile into the kohonen map. we naively assume that history repeats itself. this way all market variations with slight changes have already been in the past. and now we can pseudoscientifically determine the current state of the market. especially interesting is the trajectory of the momentum movement now using the kohonen map. maybe we will be able to switch in time between trend and flat trades.

sorry for the capital letters, sitting in a cast.

 
Mislaid:


Actually the idea was suggested to see a positive MO. The tester is no authority whatsoever. And, why race tests when you can create a tool that tells you whether to dig here or not. Also, the indicator has a basket of currencies at the input. It's practically impossible to get such a picture on any pair.

So here with the spread there is only a small positive island. A delay inside the spread really extends the positive range.

The tester is not an authority, but excel is definitely cool, yeah. Once again, the tester has a normal report, clear and familiar to all.

It's a conspiracy, first the doctor with matkad, now excel.

 

My suggestion for the branch is to take a more practical approach. Do we work by trend, Intraday or where? If trending, do we work through breakdowns? Maybe the question is: on what timeframe is it more profitable to work? The patterns of price movements can be found on different timeframes, but they will be different.

 
C-4: Ultimately it all comes down to a matter of faith. You either believe your TS and that it will work in the future, or you don't. It does not matter how you come to this belief, whether through sophisticated forward testing or simply "by eye" looking at the results on history - the bottom line is the same, it is your belief.

I somewhat disagree, as the financial series is continuous and one action flows from the other.

In my view, there should be a logical chain between past data and future data - at least there is some understanding of this relationship.

 
jelizavettka:

What's interesting, what is the exit system? Profitability 1.63, still not enough, there is room for improvement.
 
excelf:
What's interesting, what is the exit system? Profitability 1.63, still not much, there is room for improvement.
Would you like to take part in the improvement?
 

The problem has been moved from another branch:

The probability of events 3 (P3) and 4 (P4) at time t must be determined. The price went from point 0 to point 1, then rolled back to point 2.
Input data: C1, C2, C3, C4 are prices at points 1,2,3,4 respectively.



 
DmitriyN:

The problem is from another thread:

The probabilities of events 3 (P3) and 4 (P4) at time t need to be determined. The price went from point 0 to point 1, then rolled back to point 2.
Input data: C1, C2, C3, C4 are prices at points 1,2,3,4 respectively.




If C(t) has a normal distribution, as shown in the figure, then event 4 is more likely because it has a smaller deviation from 0. If the price increments have a normal distribution, then both events are equally likely if dc2-3 = dc2-4. The assumption of SB implies a normal distribution of the price increments.

This is all theory irrelevant to practice. Prices are not SBs, and neither they nor their increments have a normal distribution.

 
gpwr:
That is, you think that the probability at time t3 that price goes against the trend (from point 2 to point 4) is greater than the probability that it will go against the trend (from point 2 to point 4),
than the probability that it will go against the trend (from point 2 to point 3)?